Comparative assessment of Auto Regressive Integrated Moving Average with Explanatory variable (ARIMAX) and Neural Network Autoregressive models with Exogeneous inputs (NNARX) for forecasting the old-world bollworm, Helicoverpa armigera (Lepidoptera: Noctuidae) in India

The old world bollworm, Helicoverpa armigera Hubner (Lepidoptera: Noctuidae) is a key polyphagous agricultural pest with global wide distribution. To combat the damage caused by H. armigera farmers rely heavily on pesticides which is not a benevolent practice, environmentally and economically. To provide a more effectual and precise information on timely application of insecticides, this research was intended to develop a forecast model to predict the future trend of pod borer population by means of pheromone trap catch using Autoregressive Integrated Moving Average (ARIMAX) and Artificial Neural Networks (NNARX) with weather parameters as exogeneous variables. Several ARIMAX (p, d, q) and NNARX models were fitted by using the historical trap catch input data in different combinations. ARIMAX (2,0,0) with maximum temperature and rainfall as external variables was selected as the best ARIMAX fit. The neural network (10–32-1) was found to be the best fit to predict the male moth catches of old world bollworm from September, 2021 to August, 2023. A comparative assessment of ARIMAX and NNARX, showed that the NNARX models were found best suit for effective pest prediction to suggest timely intervention of control measures with appropriate decision-making schedule for application of insecticides